News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
The difference between AI, machine learning and deep learning
The difference between AI, machine learning and deep learning
AI (Artificial Intelligence) is the future, science fiction, a part of our daily life. All the conclusions are right, just to see what you are talking about in the end of AI.
For example, when Google DeepMind developed the AlphaGo program to defeat the Lee Se-dol of Korean professional go player, the media used DeepMind, such as AI, machine learning and deep learning, to describe the success of DeepMind. The reason AlphaGo beat Lee Se-dol, these three technologies have made a valiant record, but they are not the same.
To get to know their relationship, the most direct way of expression is concentric circles. They first presented the idea, followed by machine learning. When machine learning flourishing, they showed deep learning. Today's AI burst is driven by deep learning.
A picture of the difference between AI, machine learning and deep learning
From decline to prosperity
In 1956, the Dartmouth Conference (Dartmouth Conferences), computer scientists first proposed the term "AI", AI was born, in the subsequent days, AI laboratory became the "dream object". Over the past decades, people's perception of AI has changed from time to time. Sometimes AI is regarded as a precursor and the key to human culture in the future. Sometimes it is regarded as a mere concept of technology dregs. Ambition is too large to fail. Frankly speaking, until 2012 AI still had these two characteristics.
In the past few years, AI has been bursting, and it has been more and more rapid since 2015. The reason for the rapid development is due to the general improvement of GPU, which makes parallel disposal faster, cheaper and more powerful. Another reason is the unlimited expansion of the practical storage capacity, the large range of data generation, for example, picture, text, sale, map data information.
AI: let machines show human intelligence
Back to the summer of 1956, at that time, the AI pioneer's vision was to build a complex machine (let's just drive the computer at that time), and then let the machine show the characteristics of human intelligence.
This concept is what we call "General AI", that is to create a super machine, let it have all human perception, and even surpass human perception, it can be considered like human beings. We often see this machine in movies, like C-3PO, terminator.
One more concept is "Narrow AI". Simply speaking, "weak AI" can accomplish some specific tasks like human beings, and it may do better than human beings. For example, Pinterest services classify images with AI, and Facebook identifies faces with AI, which is "weak AI".
The above example is a case of the practice of "weak artificial intelligence", which used to show some of the characteristics of human intelligence. How do you do it? Where do these intelligence come from? With the problem, we will deepen our understanding and come to the next circle, which is the machine learning.
Machine learning: a way to reach the AI goal
Generally speaking, machine learning is to use algorithms to truly analyze data, learn from time to time, and then make a distinction and forecast for events in the world. At this point, the researchers will not write software, confirm the special instruction set, and let the program complete the special task. On the contrary, the researchers will use a lot of data and algorithms to "exercise" the machine, so that the machine can learn how to perform the task.
The concept of machine learning is proposed by early AI researchers. In the past few years, machine learning has presented many algorithms, including decision tree learning, resolution logic programming, clustering analysis (Clustering), reinforcement learning, Bayesian network and so on. As you know, no one really arrives at the ultimate goal of "strong worker intelligence". With the early machine learning method, we even fail to achieve the goal of "weak AI".
In the past many years, the best application case of machine learning is "computer vision". To complete computer vision, researchers still need to manually write large amounts of code to complete the task. The researchers manually wrote the classifier, such as edge detection filter, so long as the program could decide where the object started and where it ended, the shape detection could confirm whether the object had 8 sides; the classifier could identify the character "S-T-O-P". After manual grouping, researchers can develop algorithms to identify meaningful images, and then learn to distinguish, which is sure that it is not a stop sign.
This method can be used, but it is not very good. If it is in the fog, when the visibility of the sign is low, or a tree blocks a part of the sign, its recognition will fall. Until recently, computer vision and image detection techniques were far from human ability, because it was too easy to make mistakes.
Deep learning: technology to complete machine learning
Artificial neural network (Artificial Neural Networks) is another algorithm. It is also proposed by experts in early machine learning. It existed for decades. The idea of Neural Networks derives from our understanding of the human brain - the connections of neurons to each other. There are differences between the two. The neurons of human brain are connected by specific physical intervals. Artificial neural network has independent layers, cohesion and data transmission direction.
For example, you may draw a picture, cut it into a lot of blocks, and then implant it into the first layer of the neural network. The first layer of independent neurons will be
For example, when Google DeepMind developed the AlphaGo program to defeat the Lee Se-dol of Korean professional go player, the media used DeepMind, such as AI, machine learning and deep learning, to describe the success of DeepMind. The reason AlphaGo beat Lee Se-dol, these three technologies have made a valiant record, but they are not the same.
To get to know their relationship, the most direct way of expression is concentric circles. They first presented the idea, followed by machine learning. When machine learning flourishing, they showed deep learning. Today's AI burst is driven by deep learning.
A picture of the difference between AI, machine learning and deep learning
From decline to prosperity
In 1956, the Dartmouth Conference (Dartmouth Conferences), computer scientists first proposed the term "AI", AI was born, in the subsequent days, AI laboratory became the "dream object". Over the past decades, people's perception of AI has changed from time to time. Sometimes AI is regarded as a precursor and the key to human culture in the future. Sometimes it is regarded as a mere concept of technology dregs. Ambition is too large to fail. Frankly speaking, until 2012 AI still had these two characteristics.
In the past few years, AI has been bursting, and it has been more and more rapid since 2015. The reason for the rapid development is due to the general improvement of GPU, which makes parallel disposal faster, cheaper and more powerful. Another reason is the unlimited expansion of the practical storage capacity, the large range of data generation, for example, picture, text, sale, map data information.
AI: let machines show human intelligence
Back to the summer of 1956, at that time, the AI pioneer's vision was to build a complex machine (let's just drive the computer at that time), and then let the machine show the characteristics of human intelligence.
This concept is what we call "General AI", that is to create a super machine, let it have all human perception, and even surpass human perception, it can be considered like human beings. We often see this machine in movies, like C-3PO, terminator.
One more concept is "Narrow AI". Simply speaking, "weak AI" can accomplish some specific tasks like human beings, and it may do better than human beings. For example, Pinterest services classify images with AI, and Facebook identifies faces with AI, which is "weak AI".
The above example is a case of the practice of "weak artificial intelligence", which used to show some of the characteristics of human intelligence. How do you do it? Where do these intelligence come from? With the problem, we will deepen our understanding and come to the next circle, which is the machine learning.
Machine learning: a way to reach the AI goal
Generally speaking, machine learning is to use algorithms to truly analyze data, learn from time to time, and then make a distinction and forecast for events in the world. At this point, the researchers will not write software, confirm the special instruction set, and let the program complete the special task. On the contrary, the researchers will use a lot of data and algorithms to "exercise" the machine, so that the machine can learn how to perform the task.
The concept of machine learning is proposed by early AI researchers. In the past few years, machine learning has presented many algorithms, including decision tree learning, resolution logic programming, clustering analysis (Clustering), reinforcement learning, Bayesian network and so on. As you know, no one really arrives at the ultimate goal of "strong worker intelligence". With the early machine learning method, we even fail to achieve the goal of "weak AI".
In the past many years, the best application case of machine learning is "computer vision". To complete computer vision, researchers still need to manually write large amounts of code to complete the task. The researchers manually wrote the classifier, such as edge detection filter, so long as the program could decide where the object started and where it ended, the shape detection could confirm whether the object had 8 sides; the classifier could identify the character "S-T-O-P". After manual grouping, researchers can develop algorithms to identify meaningful images, and then learn to distinguish, which is sure that it is not a stop sign.
This method can be used, but it is not very good. If it is in the fog, when the visibility of the sign is low, or a tree blocks a part of the sign, its recognition will fall. Until recently, computer vision and image detection techniques were far from human ability, because it was too easy to make mistakes.
Deep learning: technology to complete machine learning
Artificial neural network (Artificial Neural Networks) is another algorithm. It is also proposed by experts in early machine learning. It existed for decades. The idea of Neural Networks derives from our understanding of the human brain - the connections of neurons to each other. There are differences between the two. The neurons of human brain are connected by specific physical intervals. Artificial neural network has independent layers, cohesion and data transmission direction.
For example, you may draw a picture, cut it into a lot of blocks, and then implant it into the first layer of the neural network. The first layer of independent neurons will be